Recurring issues

Open Insights → Performance → Feedback Analytics. What keeps going wrong groups selected Feedback Cases by their stored classification category; AI does not create these groups. Select a category to open its detail.

Category frequency is the number of Cases created in the selected dates and scope. The detail compares it with the immediately preceding window of identical duration. Percentage change = (current Cases − previous Cases) ÷ previous Cases. A dash means the prior count is zero, so percentage change is undefined.

The recurrence signal is narrower: for each Property, it counts a current-period Case created after an earlier Case in the same category had been resolved or closed. “Properties” counts Properties with at least one such later occurrence. This is observational evidence, not proof that the same root cause returned.

Daily trend is used for periods up to 31 days, weekly up to 120 days, and monthly for longer periods. Property rows show current cases, open cases, open High/Urgent cases and change. Resolution rate is current Resolved/Closed Cases ÷ current category Cases. Median response/resolution time excludes missing timestamps. Outcome distribution concerns current resolved/closed Cases.

Frequency is not severity. A frequent Low-priority category is not automatically more serious than a rare Urgent Case. Use Open matching cases, recent recovery actions and sanitized guest excerpts to inspect the contributing evidence. The drill-down respects accessible Properties and the active Survey/source/language filters.

Next: Feedback and recovery metrics or Feedback inbox.